The appearance of the ‘Themistocles Decree’ in 1960 incited a flurry of publications debating its authenticity. By 1962 Dow counted more than thirty.Footnote ¹ It is safe to say the debate is over. Scholars now are more likely to ask who created such a forgery and when, or why Troezenians in the third century inscribed an Athenian document pretending to be from the fifth.Footnote ² Seen from this perspective, as Johansson noted in one of the more recent papers on it, ‘[the decree] is so full of anachronistic features that it is hard to believe that anyone could ever have thought it authentic’.Footnote ³ And yet among those who thought it authentic, at least in part, were heavyweights like Lewis. His arguments for its authenticity have not endured scrutiny, but they point to an overlooked historiographic puzzle.

The puzzle comes in the decree’s final lines, just before it breaks off:

Where Hignett saw an irremediable chronological problem, Lewis found ‘the clearest single piece of evidence for authenticity’.Footnote ⁵ Observing that ancient writers commonly used the verb μεθίστημι to describe ostracism, he suggested that the language reflected ‘official and archaic’ idiom, perhaps drawing from a source from a time before ὀστρακίζω became the standard word for the practice. It seems that Lewis came to feel reservations about his argument.Footnote ⁶ And rightly, because, as Kennelly observed, μεθίστημι commonly meant ‘exile’ in fourth-century texts. If none of the sources using the term was earlier than the fourth century (including the Themistocles Decree), it is not surprising that they all used it to refer to ostracism.Footnote ⁷

Lewis’s idea that μεθίστημι was ‘official and archaic’ was based on the common belief that the exiles in the decree were the ostracized, a select group that at the time included such luminaries as Aristides and Xanthippus. According to the Athenaion Politeia, the Athenians ‘recalled all the ostracized in the archonship of Hypsichides, on account of the expedition of Xerxes’ (κατεδέξαντο πάντας τοὺς ὠστρακισμένους ἄρχοντος Ὑψηχίδου, διὰ τὴν Ξέρξου στρατείαν, 22.8). With this passage in mind, Lewis was right to puzzle over the decree’s language. He reasoned, if the exiles in the Themistocles decree were the ostracized, and its author did not call them that, he ‘would have shown remarkable restraint and knowledge’. The odd language thus had to be a sign of authenticity.Footnote ⁸ These days most scholars do not believe the document is authentic. But the oddity remains. Why did its author write ‘the exiles for the ten years’ and not simply ‘the ostracized’, which was the standard expression in the third, fourth or even fifth centuries?

The simplest reason its author did not write ‘the ostracized’ is because the small group of individuals ostracized by that point were not the people he had in mind. Ignoring the context provided by the Ath. Pol. passage for the moment (we will return to it later), it stretches the language of the Troezen decree to take τοὺς μὲν μεθεστηκότας τὰ [δέκα] ἔτη to refer to the ostracized. In fourth-century texts μεθίστημι can refer to the legal penalty of exile, but that is not its basic meaning. μεθίστημι means ‘to change’, ‘to move’ or ‘to remove’ (LSJ). It can also refer to someone who has simply left a room (for example, Aeschin. 3.117). Or it might denote someone who has moved metaphorically, that is, switched allegiance (for example, Thuc. 8.15.1; Xen. Hell. 1.4.9). It is impossible to determine its nuance in our passage by looking at it in isolation. Context and related passages are important. Here are some that Lewis thought supported his idea:

When we look at these passages, it is understandable why Lewis sensed a strong connection between μεθίστημι and ostracism. But there are key differences with our passage.

First, our passage uses the perfect. The semantics of the perfect are a notorious problem in Greek grammar. Some intransitive forms can be active but feel passive, as in our case; others can connect a present state to a past event tightly (the ‘resultative’), or loosely (Schwyzer’s ‘attained state’) or not at all (the ‘iterative/intensive’).Footnote ⁹ Aided by digital tools, recent work in historical linguistics has improved our understanding of the tense and how it changed over time.Footnote ¹⁰ μι-verbs like ἵστημι and its compounds are also a special case. As shown in the examples above, μεθίστημι can take a direct object, but it is not obligatory. Its transitive form is in semantic terms an ‘accomplishment’ verb. But in the perfect, as Crellin has shown, ἵστημι undergoes ‘causative alternation’ and becomes a ‘COS [change-of-state] verb’. Such verbs describe a change of state in the subject; these are also known as ‘unaccusatives’. It is a mistake to simply take a transitive connotation, make it perfect and assume that it will carry the nuance of the English word.

Jameson’s translation of τοὺς μὲν μεθεστηκότας τὰ [δέκα] ἔτη as ‘those who have been sent into exile for ten years’ can refer to the ostracized only thanks to an ambiguity in the translation. The English expression ‘for X time’ can convey intention as well as duration. For example, I can say ‘I have leased this car for five years’ or ‘I have left home for a week’ before the stated interval has ended.Footnote ¹¹ The ambiguity is possible only with verbs that connote a decision or judgment. I cannot say ‘I have travelled for a week’ before the end of the week. But I can say ‘I have been banished for ten years’ before the ten years are up. This is the only way Jameson’s translation can refer to the ostracized, because the first ostracism was less than ten years before.

It is unlikely that Greek οἱ μεθεστηκότες τὰ [δέκα] ἔτη can convey that nuance. The most natural way to read the perfect with a simple accusative of time is as referring to a full or complete interval.Footnote ¹² To mark the time interval as one intended rather than completed, our author would have had to phrase it differently; say, οἱ μετασταθέντες ἐπὶ τὰ δέκα ἔτη, or εἰς τὰ δέκα ἔτη; or simply, οἱ ἐξωστρακισμένοι.Footnote ¹³ Furthermore, Jameson’s translation foregrounds the anterior event that led to the exile, not the state of exile itself. As Crellin shows, perfect ‘COS verbs’ like ἀναχωρέω, ἐπιβαίνω, φύω or ἵστημι semantically tend to foreground the present state of the subject, rather than the anterior event that led to it.Footnote ¹⁴ Other euphemisms of banishment, like φεύγω or ἐξέρχομαι, also fall into this category. Replacing μεθεστηκότας with the synonymous ἐξελελυθότας should make the problem clear.Footnote ¹⁵ The Greek can only be saying, ‘those who have been away for ten years’. This may not apply to someone who has returned from their ten-year exile early.

In short, the ambiguity that allows Jameson’s translation to refer to the ostracized is not present in the original. ‘Ten years’ must refer to an interval that has culminated, not one that was intended when each person was individually sentenced to a ten-year banishment. On the other hand, the Greek perfect here is ambiguous about whether, for our author, the decade culminated at some point before the decree’s dramatic present, or in it.Footnote ¹⁶ A small detail in his language might shed some light here. Jameson left out ‘the’ from his translation, but that word might hold the key to what the author had in mind.

According to Kühner–Gerth (§465.13a), definite cardinal-numeric substantives, like τὰ δέκα ἔτη, might refer to previously mentioned indefinite substantives, or anticipate subsequent specification, or be themselves specific, distributive or summative, or oppose a total to a part. None of these readily apply to our passage as it has been interpreted. Plutarch also does not use the definite article in a passage closely related to ours, Them. 11.1:Footnote ¹⁷

It is surprising that Plutarch, like the author of the decree, also did not call the exiles ‘the ostracized’, even though he suggests that the provision was intended to secure the return of the ostracized Aristides. He believed Aristides had been away for three years (Arist. 8.1). Perhaps that is why he did not write ‘ten years’. But the expression he uses instead, τοῖς ἐπὶ χρόνῳ μεθεστῶσιν, is unusual; editors have long eyed it with suspicion.Footnote ¹⁸ If his source stated something like what we read in the Decree, the phrase ‘for a time’ might represent Plutarch’s own gloss on ‘the ten years’. His reading allows him to include his hero under the recall without associating him too closely with ‘those who have been exiled for the ten years’.Footnote ¹⁹ Note his use of the short perfect of the verb. As Crellin suggests, in verbs with two perfect active stems like ἵστημι, the kappa-form tends to be ‘causative’, implying the subject’s responsibility for its present state, whereas the short stem functions as ‘anticausative’.Footnote ²⁰ In contrast to how the Decree describes the exiles, Plutarch’s language seems to reflect his belief in Aristides’ guiltlessness. But the language of the Decree provides no reason to think that its author had Aristides in mind at all.Footnote ²¹

Let us return to the Decree’s ‘the ten years’. I suggested above that none of the normal uses of the definite article with cardinal-numeric substantives applies readily to our passage. An idiomatic use of the definite article that does apply is its ability to refer to something mentioned previously or generally known, like Latin ille or iste (K.–G. §461.8; CGCG 28.5). Thucydides (1.11) provides an instructive example with the same phrase:

Thucydides did not have to explain that in the tenth year of the war Troy fell. The definite article underscores the common familiarity of this fact. Similarly, when our author wrote ‘the ten years’ rather than simply ‘ten years’, he must have been referring to a ten-year period as familiar to him as ‘the ten years’ of the Trojan War. If this was not the familiar term of ostracism, for the reasons I provided above, there is only one other possibility that would fit the context. Again, Thucydides points the way (1.18.1–2):

The notion of the ‘fall of the tyrants of Greece’ is chronologically indeterminate, as is ‘not many years’. But ‘ten years’ is the traditional, specific interval between Marathon and Salamis, and a cornerstone of Athenian chronology.Footnote ²² The Marmor Parium shows that chronographers’ knowledge of the Athenian archon-list before Salamis was spotty. But its author knew that precisely ten archon-years separated Salamis from Marathon (BNJ 239 a48, a51). This interval was familiar not just to historians. Lysias has Xerxes arrive ‘in the tenth year’ after Marathon (2.27). Plato calculates in the opposite direction: ‘nearly ten years before Salamis, Datis arrived’ (Leg. 3.698c). Our author’s familiarity with the decade as a definite chronological era in Athenian history might have lulled him into using the definite article without realizing it was slightly premature to have Themistocles refer to ‘the’ ten years before their defining event had supposedly occurred.Footnote ²³

Whoever the exiles in the author’s mind, his language precludes the possibility that they were the ostracized. They have not been sentenced to individual exiles for ‘ten years’ that are ongoing. Instead, they have been collectively removed from Athens for ‘the ten years’ that have passed at the decree’s dramatic present.Footnote ²⁴ But if they were not the ostracized, who did our author think they were? To answer this question we must enter the realm of speculation, but the text behind the exiles’ identification with the ostracized might point to a different possibility.

Ath. Pol. 22 traces the practice of ostracism against the growth of the democracy, from the practice’s invention by Cleisthenes (1), to its first use in the third year after Marathon (3), to the recall of all the ostracized with the invasion of Xerxes (8). Most scholars believe this chapter’s elegant chronological synchronisms owe a special debt to Androtion. One of his fragments (FGrHist 324 F 6) is tantalizingly close to the Ath. Pol. text, although it is uncertain if the fragment is slightly corrupt, if Ath. Pol. wanted to correct him or if something else is going on.Footnote ²⁵ How these texts relate to the well-worn question about the date of ostracism’s introduction does not concern us here. What does is the fact that both texts agree that suspicion of the ‘friends of the tyrants’ was behind it. Philochorus also apparently shared this view (FGrHist 328 F 30).Footnote ²⁶ As Harding notes, ‘we do not have any extant fragments from the Atthidographers directly related to the expulsion of the Peisistratidai’.Footnote ²⁷ It seems clear, however, that they were quite interested in their ‘friends’. And they were convinced that they continued to pose a problem after the tyrants were expelled (cf. Ath. Pol. 19.6–20.1 with Hdt. 5.65–6).

Jacoby showed that for the period from Pisistratus to Xerxes the Atthidographers not only ‘followed [Herodotus] in the main lines’ but also tried to supplement or even correct him.Footnote ²⁸ For example, Herodotus’ memorable story about Pisistratus’ first return gave rise to much speculation about the status and identity of the woman who accompanied him.Footnote ²⁹ It is futile to guess why Cleidemus thought Phye married Hipparchus and her father was named Socrates; or why he thought that Charmus’ daughter married Hippias. But the inference that he had been Hippias’ lover was almost certainly based on an inscribed altar that Charmus once dedicated to the god Eros.Footnote ³⁰ Curiously, while Cleidemus makes Charmus Hippias’ erastês, Plutarch believed he was Pisistratus’ erômenos (Sol. 1.4). The difference might indicate a chronological dispute.Footnote ³¹ We have no way of knowing how, if at all, his name’s prominence on a bronze stele on the Acropolis for ‘accursed traitors’ contributed to the notion that Charmus’ son Hipparchus led the ‘tyrannist’ faction when he was ostracized (Lycurg. Leoc. 117–18; Ath. Pol. 22.4).Footnote ³²

The Atthidographers used Herodotus similarly for the Persian Wars, drawing on him ‘in the main lines’ while also supplementing or correcting him in the margins. In a revealing fragment, Cleidemus assigns to the tribe Aiantis the fifty-two deaths at Plataea that Herodotus assigned to the entire Athenian contingent (FGrHist 323 F 22; cf. Hdt. 9.70.5). Whatever the meaning of his divergence from Herodotus, Cleidemus possibly based his opinion on a document or an oral tradition (or both) about an Aiantid sacrifice to the Nymphs.Footnote ³³ No doubt he and other writers had similar reasons to dispute how the Athenians funded their navy when they evacuated the city (FGrHist 323 F 21; Ath. Pol. 23. 1; cf. Hdt. 8. 41); how many ships the Naxians sent to Salamis and how they fared (Hellanicus FGrHist 323a F 28; cf. Hdt. 8.46.3); or where Xerxes sat to watch the battle (Phanodemus FGrHist 325 F 24; Aristodemus BNJ 104 F 1, 1.2; cf. Hdt. 8.90.4).

The Themistocles Decree was inscribed in the third century, but a plausible case can be made that it derives from this post-Herodotean historiographic tradition. Cleidemus himself has been proposed as its author.Footnote ³⁴ We know he credited Themistocles with organizing the evacuation of the city, just as the Decree does. And Plutarch mentions him in a passage that overlaps the Decree (Them. 10.2–11.1, at 10.4). This does not prove Cleidemus composed the text, of course. Someone else could have created it based on his version, or on a related one. But whoever wrote it put a much more patriotic spin on the events than his ultimate source did. Herodotus presented the evacuation of Athens as a desperate measure after Artemisium (8.40–1). The Decree’s author believed that Themistocles knew already before Artemisium where the key battles would take place and masterminded the strategy accordingly. In his version, Athenians did not panic and did not need their allies’ help to evacuate. Instead, they organized themselves calmly, collectively and democratically to face the threat, passing a formal decree in the assembly and distributing crews to their ships by lot. Perhaps the Miltiades Decree that we hear about would have similarly ‘corrected’ Herodotus’ account of Marathon.Footnote ³⁵

I propose that the provision about exiles in the Themistocles Decree draws on, and against, Herodotus in a similar fashion. While he had little to say about the tyrants’ Athenian allies, they appear in his text at key moments. We read about nameless παραστάται or ‘comrades’ (LSJ) beside Hippias on the field of Marathon (6.102, 107.4). After the battle, we read about pro-tyrant conspirators raising a shield-signal from within the ranks of the victorious side (6.121–4). Most relevantly, we also read about anonymous ‘Pisistratids’ accompanying Xerxes—ten years later (7.6). When the Persians reached Athens, Herodotus says they tried to convince the Athenians to surrender the Acropolis (8.52.2). The day after it fell, one day before the battle, these ‘exiles of the Athenians’, οἱ φυγάδες τῶν Ἀθηναίων, visited it again and bore witness to the burnt olive-tree sprouting again, an omen of Athenian resilience (8.54–5).

These glimpses of Athenian traitors in Herodotus might not seem particularly edifying to us, but his Atthidographic successors will have found them an irresistible loose end. We know they read Herodotus’ text carefully and tried to supplement and correct him. We know they thought that ‘Pisistratid friends’ continued to be a problem long after the tyrants were expelled and speculated about their identity and fate. Therefore, it stands to reason that Herodotus’ references to ‘Pisistratid exiles’ following Xerxes raised the question of what ultimately became of them. The end of the ‘Themistocles Decree’ might hint at their answer. Depending on how closely the author of the Decree followed Herodotus, we can imagine that the omen they saw on the Acropolis might have spurred the exiles to cross over to Salamis and rejoin the Athenians just in time for the battle.

If this argument is right it cannot shed light on the historicity of the exiles or their recall, one way or the other. From top to bottom the Decree is probably a product of fourth-century historiographic imagination. But it is also possible—as Lewis and other defenders of its authenticity insisted—that its author drew on a document or an oral tradition independent of Herodotus. He might have had good reason to believe in an exiles’ kathodos in the days of the Persian Wars and to try to reconcile it with Herodotus’ narrative.Footnote ³⁶

Whatever the truth behind these events, it is clear that whoever wrote the Decree believed that Xerxes’ invasion prompted the democracy to extend an olive-branch to its exiles. I have argued that these included a dangerous group whom he described as ‘the exiles for the ten years’. Unlike the ostracized, they had shown themselves ready to ally with the Persians.Footnote ³⁷ He imagined Themistocles inviting them to return in a spirit of homonoia, while acknowledging the final decision to take them back was in the hands of the demos.